Objective: The artificial pancreas (AP) has shown promising potential in achieving closed-loop glucose control for individuals with type 1 diabetes mellitus (T1DM). However, designing an effective control policy for the AP remains challenging due to the complex physiological processes, delayed insulin response, and inaccurate glucose measurements. While model predictive control (MPC) offers safety and stability through the dynamic model and safety constraints, it lacks individualization and is adversely affected by unannounced meals. Conversely, deep reinforcement learning (DRL) provides personalized and adaptive strategies but faces challenges with distribution shifts and substantial data requirements.
Methods: We propose a hybrid control policy for the artificial pancreas (HyCPAP) to address the above challenges. HyCPAP combines an MPC policy with an ensemble DRL policy, leveraging the strengths of both policies while compensating for their respective limitations. To facilitate faster deployment of AP systems in real-world settings, we further incorporate meta-learning techniques into HyCPAP, leveraging previous experience and patient-shared knowledge to enable fast adaptation to new patients with limited available data.
Results: We conduct extensive experiments using the FDA-accepted UVA/Padova T1DM simulator across three scenarios. Our approaches achieve the highest percentage of time spent in the desired euglycemic range and the lowest occurrences of hypoglycemia.
Conclusion: The results clearly demonstrate the superiority of our methods for closed-loop glucose management in individuals with T1DM.
Significance: The study presents novel control policies for AP systems, affirming the great potential of proposed methods for efficient closed-loop glucose control.
Objective: Navigation through tortuous and deformable vessels using catheters with limited steering capability underscores the need for reliable path planning. State-of-the-art path planners do not fully account for the deformable nature of the environment.
Methods: This work proposes a robust path planner via a learning from demonstrations method, named Curriculum Generative Adversarial Imitation Learning (C-GAIL). This path planning framework takes into account the interaction between steerable catheters and vessel walls and the deformable property of vessels.
Results: In-silico comparative experiments show that the proposed network achieves a 38% higher success rate in static environments and 17% higher in dynamic environments compared to a state-of-the-art approach based on GAIL. In-vitro validation experiments indicate that the path generated by the proposed C-GAIL path planner achieves a targeting error of 1.26 ±0.55mm and a tracking error of 5.18 ±3.48mm. These results represent improvements of 41% and 40% over the conventional centerline-following technique for targeting error and tracking error, respectively.
Conclusion: The proposed C-GAIL path planner outperforms the state-of-the-art GAIL approach. The in-vitro validation experiments demonstrate that the path generated by the proposed C-GAIL path planner aligns better with the actual steering capability of the pneumatic artificial muscle-driven catheter utilized in this study. Therefore, the proposed approach can provide enhanced support to the user in navigating the catheter towards the target with greater accuracy, effectively meeting clinical accuracy requirements.
Significance: The proposed path planning framework exhibits superior performance in managing uncertainty associated with vessel deformation, thereby resulting in lower tracking errors.
Data-driven methods for lesion generation are quickly emerging due to the need for realistic imaging targets for image quality assessment and virtual clinical trials. We proposed a generative adversarial network (GAN) architecture for conditional generation of lung lesions based on user-specified classes of lesion size and solidity. The network consists of two discriminators, one for volumetric lesion data, and one for radiomics features derived from the lesion volume. A Wasserstein loss with gradient penalty was adopted for each discriminator. Training data were drawn from contoured and annotated lesions from a public lung CT database. Four quantitative evaluation methods were devised to assess the network performance: 1) overfitting (similarity between generated and real lesions), 2) diversity (similarity among generated lesions), 3) conditional consistency (capability of generating lesions according to user-specified classes), and 4) similarity in distributions of various lesion properties between the generated and real lesions. Ablation studies were also performed to investigate the importance of individual network component. The proposed network was found to generate lesions that resemble real lesions by visual inspection. Solid lesions are distinct from non-solid ones, and lesion sizes largely correspond to their specified classes. With a classifier trained on real lesions, the classification accuracies of generated and real lesions in both solid and non-solid classes are similar. Radiomics features of generated and real lesions were found to have similar distributions, indicated by the relatively low Kullback-Leibler (KL) divergence values. Furthermore, the correlations between pairwise radiomics features in generated lesions were comparable to those of real lesions. The proposed network presents a promising approach for generating realistic lesions with clinically relevant features crucial for the comprehensive assessment of medical imaging systems.
Objective: This study investigates relationships between the oscillation behavior of the medial and superior vocal fold (VF) surfaces during sustained phonation in a human cadaver hemilarynx.
Methods: An experimental test stand synchronously captured the medial and superior VF surfaces of a human ex vivo hemilarynx during sustained phonation using two high-speed camera setups in 24 experimental settings. The 3D coordinates of the medial VF surface were reconstructed by triangulation of sewn-in marker points, while laser-based reconstruction was used for the superior VF surface. Correlation analysis and linear regression were used to quantify the connections of the mean and maximal vertical and lateral VF displacements and the VF velocities. Additionally, stepwise linear regression was used to analyze the impact of the measurement variables mean flow rate, adduction and elongation.
Results: Strong linear relationships between all of the tested corresponding parameter pairs of the superior and medial VF surfaces were found (p<.001). Mean and maximum vertical displacements of the medial surface were both approximately 50% of the superior surface. The mean lateral displacements for the medial surface were 12% below the superior surface but 12% higher for the maximum values. The mean and maximum VF velocities were 32% and 36% lower for the medial surface.
Conclusion: The suggested multi-modal test stand allows efficient, comprehensive analysis of human hemilarynges and provides promising information about the interaction of the different VF areas and opens up the systematic analysis of multiple hemilarynges.
Significance: In future, our results could integrate into ENT diagnostics using 3D laryngoscopy where the hidden medial VF surface dynamics may be predicted from the observable superior surface.
Objective: Teleoperated robot-assisted minimally-invasive surgery (RAMIS) offers many advantages over open surgery, but RAMIS training still requires optimization. Existing motor learning theories could improve RAMIS training. However, there is a gap between current knowledge based on simple movements and training approaches required for the more complicated work of RAMIS surgeons. Here, we studied how surgeons cope with time-dependent perturbations.
Methods: We used the da Vinci Research Kit and investigated the effect of time-dependent force and motion perturbations on learning a circular pattern-cutting surgical task. Fifty-four participants were assigned to two experiments, with two groups for each: a control group trained without perturbations and an experimental group trained with 1Hz perturbations. In the first experiment, force perturbations alternatingly pushed participants' hands inwards and outwards in the radial direction. In the second experiment, the perturbation constituted a periodic up-and-down motion of the task platform.
Results: Participants trained with perturbations learned how to overcome them and improve their performances during training without impairing them after the perturbations were removed. Moreover, training with motion perturbations provided participants with an advantage when encountering the same or other perturbations after training, compared to training without perturbations.
Conclusion: Periodic perturbations can enhance RAMIS training without impeding the learning of the perturbed task.
Significance: Our results demonstrate that using challenging training tasks that include perturbations can better prepare surgical trainees for the dynamic environment they will face with patients in the operating room.
Background: Strength-duration (SD) curve, rheobase and chronaxie parameters provide insights about neural activation dynamics and interdependence between pulse amplitude and duration, for diagnostics and therapeutic applications. The existing SD curve estimation methods are based on open-loop uniform and/or random pulse durations, which are chosen without feedback from neuronal data.
Objective: To develop a method for closed-loop estimation of the SD curve, where the pulse durations are adjusted iteratively using the neuronal data.
Method: In the proposed method, after the selection of each pulse duration through Fisher information matrix (FIM) optimization, the corresponding motor threshold (MT) is computed, the SD curve estimation is updated, and the process continues until satisfaction of a stopping rule based on the successive convergence of the SD curve parameters. The results are compared with various iterative uniform and random sampling techniques, where the SD curve estimation is updated after each sample.
Results: 250 simulation cases were run. The FIM method satisfied the stopping rule in 225 (90%) runs, and estimated the rheobase (chronaxie in parenthesis) with an average absolute relative error (ARE) of 1.57% (2.15%), with an average of 85 samples. At the FIM termination sample, methods with two and all random pulse durations, and uniform methods with descending, ascending and random orders led to 5.69% (20.09%), 2.22% (3.93%), 7.34% (40.90%), 3.10% (4.44%), and 2.05% (3.45%) AREs. In all 250 runs, the FIM method has chosen the minimum and maximum pulse durations as the optimal pulse durations for the SD curve estimation.
Conclusions: As proposed by the FIM method, the SD curve is identifiable by fitting to the data of the minimum and maximum pulse durations. However, the range of pulse duration should cover the vertical and horizontal parts of the SD curve. Also, iterative random or uniform samples from only the vertical or horizontal areas of the curve might not result in satisfactory estimation.
Significance: This paper provides insights about pulse durations selection for closed-loop and open-loop SD curve estimation.
Objective: Measurement of loop gain in patients with obstructive sleep apnea (OSA) is of great importance for determining the cause of OSA and realizing precision medicine. In this study, automatic measurement of loop gain is carried out during wakefulness based on the pseudorandom binary sequence (PRBS) CO2 stimulation method.
Methods: A respiratory data acquisition platform is designed and constructed to automate the processes of gas configuration, ventilatory stimulation with CO2 and data acquisition. The respiratory data are substituted into the model of the ventilatory control system for parameter identification, and the loop gain values are calculated and then compared with the apnea-hypopnea index (AHI) measured in a hypoxia chamber.
Results: A positive correlation is found between loop gain and AHI measured in the hypoxia chamber, with the linear correlation coefficient of approximately 0.65.
Conclusion: The feasibility of automatic measurement of loop gain using the respiratory data acquisition platform based on the PRBS CO2 stimulation method is validated, and the measured loop gain values can be used to assess the stability of ventilatory control.
Significance: This study provides an automated, rapid, and instrumented solution for loop gain measurement, laying the foundation for wide-scale clinical application of the PRBS CO2 stimulation method.
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：